A disparity range estimation technique for stereo-video streaming applications

In this paper, we propose a robust and efficient technique for frame-by-frame estimation of disparity ranges in stereo video. The proposed technique utilizes a single-layer image size reduction for faster processing and more effective noise handling. Furthermore, it applies spatial-domain non-linear filtering of both disparity and confidence maps for additional noise suppression and improved range estimation. A mechanism for supporting temporal consistency is proposed as well. Performance comparisons with recent approaches demonstrate the advantages of the proposed approach.